<Nadrieril>
Are you trying to build (import release.nix).build or (import release.nix).build.x86_64-linux?
mizu_no_oto has joined #nixos
<gchristensen>
yep, I use (import ./path/to/nixops/release.nix {}).build.x86_64-linux
kampfschlaefer has quit [(Ping timeout: 256 seconds)]
CodeWarrior has joined #nixos
xwvvvvwx has quit [(Ping timeout: 245 seconds)]
c74d has joined #nixos
matthewbauer has joined #nixos
<nekroze>
Nadrieril: gchristensen: sorry I got pulled away for a bit. I have been using callPackage so I guess I am doing it all wrong lol. Will try that line gchristensen provided
athan has quit [(Remote host closed the connection)]
<nekroze>
that worked, much appreciated
nekroze has quit [(Quit: Lost terminal)]
matthewbauer has quit [(Ping timeout: 256 seconds)]
justan0theruser has joined #nixos
c74d has quit [(Ping timeout: 258 seconds)]
justanotheruser has quit [(Ping timeout: 255 seconds)]
mzarella has joined #nixos
target_i has quit [(Quit: leaving)]
<NixOS_GitHub>
[nixpkgs] NeQuissimus opened pull request #22040: zulu: init at 8.19.0.1 (master...zulu) https://git.io/vM7ru
faffolter has quit [(Remote host closed the connection)]
markus1199 has joined #nixos
markus1189 has quit [(Ping timeout: 240 seconds)]
rejuvyesh has joined #nixos
<Judson>
I'm having continuing trouble with working on a Ruby gem with bundix, and not sure how to debug or report errors...
<Judson>
It seems like the use of the Gemfile "gemspec" directive is a problem?
<Judson>
Oh. Nothing like trying to assemble a problem description to put a finer point on the problem.
<Judson>
Issue is that bundix doesn't support path sources. Does anyone have an approach for this already?
SovereignBleak has quit [(Remote host closed the connection)]
matthewbauer has quit [(Ping timeout: 258 seconds)]
rejuvyesh has joined #nixos
<moet>
what controls the size of my runtime directory?
<moet>
/run/user/1000
<moet>
it seems to be 100 mb right now
matthewbauer has joined #nixos
mojjo has quit [(Ping timeout: 276 seconds)]
matthewbauer has quit [(Ping timeout: 247 seconds)]
<stepcut>
if I set, NIX_BUILD_HOOK=build-remote.pl, will nix only build on the remote systems in /etc/nix/machines, or will it also build locally?
nekroze has joined #nixos
nekroze has quit [(Changing host)]
nekroze has joined #nixos
mrkgnao has quit [(Ping timeout: 260 seconds)]
<nekroze>
I have need for a caching nix store but I don't want to have hydra running and building all thing all the time. Is it possible to set up some kind of cache server that stores derviations as the machines that use it need them?
<gchristensen>
well how many machines are we talking?
erictapen has joined #nixos
<nekroze>
one
<erictapen>
hello, I want to just use gnome-terminal, without the gnome desktop. Is there a way to install just this single application?
<nekroze>
it is for my libvirt machine, when I am staging my libvirt host I use nested KVM but when I rebuild the host vm, then it has to download all the packages again for the guests because its nix store was wiped when it was destroyed
<nekroze>
it makes testing take hours when I have to destroy the host and redploy the guest vms under the host vm
<erictapen>
in general I'm a bit confused about how to find out the package name corresponding to a certain application. as far as I can tell, the name in the mkDerivation isn't right
<nekroze>
but if I had a binary cache with the derviates cached locally on my KVM host that the staged host vm runs on then redeploying should be fast and I can get on with nix rather then watching it redownload 500 gem packages
filterfish has joined #nixos
<erictapen>
Would be awesome if someone could help me out
mrkgnao has joined #nixos
<nekroze>
erictapen: You can run `nix-env -qaP | grep gnome | grep terminal` bit I would recommend first installing nox `nix-env -i nox` then use that `nox gnome-terminal` which will list the options out like yaourt on arch linux
<nekroze>
all of these methods would work. rmemeber though to install it via the attribute name you must use `nix-env -iA nixos.gnome3.gnome_terminal` as opposed to using the package name such as `nix-env -i gnome-terminal`
<gchristensen>
erictapen: in this case, you would find nixos.gnome3.gnome_terminal in the output which to install you could do nix-env -iA nixos.gnome3.gnome_terminal or nix-env -iA gnome3.gnome_terminal
systemfault has joined #nixos
<gchristensen>
nekroze: (1) you may use nix-copy-closure from the host to your libvirt machine, (2) nixops should handle this for you I think
<erictapen>
ok thanks alot, forgot to say that I want to do it in a declarative way on nixos. Never got really in touch with nix-env because I was so focused on my configuration.nix. Definitely will need to have a better look into this. The name [pkgs.]nixos.gnome3.gnome_terminal still doesn't work in environment.systemPackages. Did I misunderstand the concept?
<gchristensen>
erictapen: when using pkgs directly, it would be pkgs.gnome3.gnome_terminal
<nekroze>
gchristensen: nixops does handle it but the middle host's /nix store is non persistent. so I have debian->nixos->nixos the debian box is my physical machine, the middle nixos is a libvirt nixops deployed vm running libvirt inside of that, and the final nixos machine is a guest VM deployed with nixops running on the middle host. Because the middle host has its hdd wiped when I do a nixops destroy on my
<nekroze>
debian machine I have to wait hours for that third vm to redownload all of its packages into the middle vm's nix store
<gchristensen>
nekroze: yikes, well, you could still nix-copy-closure from the middle box to the debian box and then copy it back
<nekroze>
erictapen: `environment.systemPackages = [pkgs.gnome3.gnome_terminal]` will work, and the top line of the file should have pkgs in it like so `{config, pkgs, ...}:`
<erictapen>
nekroze: nice, thats what I was looking for. Will use nix-env -qaP in the future for this. Thank you
<nekroze>
gchristensen: interesting, that might work if I set up keys in the nested host for ssh but I think it will generate a new key each redeployment of it... but that is a start
<nekroze>
erictapen: np, enjoy
<nekroze>
How does garbage collection work when copying a nix closure? if its just in the nix store wouldn't it be gotten rid of next gc?
<gchristensen>
it could be yes
<gchristensen>
you could also use nix-push which will literally make a binary cache
<erictapen>
cd niksnut
<gchristensen>
hm?
<nekroze>
gchristensen: never looked at nix-push or nix-pull. will do some reading thanks
moet has quit [(Quit: oops)]
<gchristensen>
you're welcome
mizu_no_oto has quit [(Quit: ["Textual IRC Client: www.textualapp.com"])]
<kier>
but it can't find the sshfs binary, even with sshfs-fuse added to environment.systemPackages
<kier>
ah, right now i'm just trying to get it to mount in the first place, let alone automount :p
<gchristensen>
simpson: I'm impressed by tahoe :)
erictapen has left #nixos ["Leaving"]
<simpson>
gchristensen: It's like the coolest thing that nobody knows about.
<gchristensen>
simpson: do you have recommended needed / happy / total settings?
<simpson>
gchristensen: 3/7/10 is still solid and it's what I use all the time.
pradeepch has joined #nixos
mbrgm has quit [(Ping timeout: 255 seconds)]
nekroze has quit [(Quit: Lost terminal)]
matthewbauer has joined #nixos
mrkgnao has joined #nixos
Judson has quit [(Remote host closed the connection)]
mbrgm has joined #nixos
matthewbauer has quit [(Ping timeout: 260 seconds)]
athan has joined #nixos
<gchristensen>
simpson: the nixos module says furls should be secret. I can't find further information about that. your furls are public. can you tell me about that?
hexagoxel has quit [(Ping timeout: 245 seconds)]
matthewbauer has joined #nixos
feepo has quit [(Quit: Connection closed for inactivity)]
<clever>
which one was the new version, npm2nix or node2nix?
<gchristensen>
node2nix is new I think
matthewbauer has quit [(Ping timeout: 256 seconds)]
<gchristensen>
??? the tahoe service seems very confusing to me, maybe because I don't understand
hexagoxel has joined #nixos
<simpson>
gchristensen: Anybody with a furl can wield it. So you should keep it secret if you don't want people to be able to use your stuff.
<gchristensen>
ie: write, and read what they write, but not read what I write
<simpson>
An introducer furl lets you connect to any storage on that introducer's grid.
<gchristensen>
ah
<simpson>
When we say "secret", what we really mean is that these are *capabilities*; they are raw power and it's up to their owners to safeguard them.
<gchristensen>
simpson: do you use the nixos service on clients? or via `tahoe create-client`
<simpson>
gchristensen: Depends. On my personal machines, I prefer the latter, for development and hacking.
<gchristensen>
ah
<simpson>
But in the cloud, I use the NixOS service to set up machine-wide clients with limited authority.
pie_ has quit [(Ping timeout: 264 seconds)]
slack1256 has joined #nixos
Lowl3v3l has quit [(Remote host closed the connection)]
<moet>
i got a kernel "BUG unable to handle NULL pointer dereference" just after systemd "Reached target Sleep" ..
<moet>
it happened twice today
<moet>
just wondering if there's some recommended way to start tracking that down, or reporting it upstream or something..
ckauhaus has joined #nixos
<Mic92>
moet: from application this is comming from?
MP2E has quit [(Quit: brb)]
johann__ has joined #nixos
MP2E has joined #nixos
Avaq has joined #nixos
johann__ has quit [(Client Quit)]
matthewbauer has joined #nixos
jensens has quit [(Ping timeout: 255 seconds)]
<Avaq>
Hello. Does anyone know if there is a way I can load a whole derivation -including all of the environment it uses- into my config? Eg. `services.mongodb.package = (import my/pkgs/top-level/all-packages.nix).mongodb`? I want to build a package from `master` and use it like that in my own `16.03` environment, but if I use `callPackage`, the wrong envionment is provided to the `default.nix` function.
matthewbauer has quit [(Ping timeout: 248 seconds)]
teknico has joined #nixos
<Avaq>
If I `cd my/; nix-build -A mongodb`, the build completes succesfully. But if I try to add it to my `nix/configuration.nix` like that, the build keeps failing due to being given outdated dependencies, and it forces me to copy more and more of the code in `my/pkgs/top-level` and inline it in the second argument to `pkgs.callPackage`.
<moet>
Mic92: teh kernel
<Avaq>
Every time I do, the build makes it further, but fails somewhere later down the line. It's a painful debugging experience when I have to wait two hours for the effects of every change.
<moet>
journalctl -b -1
acowley_away has quit [(Ping timeout: 245 seconds)]
<Mic92>
moet: if you have a stack trace, you can post it on a kernel bug tracker.
<shanemikel>
okay, i'll take a look tomorrow.. the best way in Bash to guarantee cleanup is to use traps, but it gets a little verbose and I can't stand Bash "arrays" so I don't use them often
matthewbauer has quit [(Ping timeout: 240 seconds)]
<shanemikel>
oh.. I see you did use a trap
cutuchiqueno has joined #nixos
<cutuchiqueno>
some time ago I played around with NixOS on an old Laptop. Today I recognized that I could not remember root passwod of the install (yeah, I know. But was just for playing). I remember from Arch that it is possible to append `init=/bin/bash` to the kernel line in GRUB2, boot and reset by `passwd`. However in NixOS this makes the kernel panic.
<cutuchiqueno>
any options or do I need a reinstall?
<lassulus>
cutuchiqueno: if you have a lifestick you can boot into it, mount the root partition and edit the hash in /etc/shadow manually
<goibhniu>
cutuchiqueno: there's a way to do it from the emergency shell too ... I think it's in the wiki, one sec
<NixOS_GitHub>
[nixpkgs] vdemeester opened pull request #22048: Fixing the wrong Git Commit hash in docker version (master...fix-docker-version-git-commit) https://git.io/vM5J2
* goibhniu
vaguely remembers also calling passwd directly from the emergency shell by using the full nix store path
<cutuchiqueno>
goibhniu: thx, this for the page I was googling for but did not find
shanemikel has quit [(Quit: My MacBook Air has gone to sleep. ZZZzzz…)]
deepfire has joined #nixos
<goibhniu>
np, I'm not sure it helps you much though ...
mudri has quit [(Ping timeout: 240 seconds)]
jgertm has joined #nixos
<cutuchiqueno>
well I am in now at least
<cutuchiqueno>
though, when I try to edit /etc/shadow/ with nano it says `Error opening terminal: unknown`
<goibhniu>
promising!
<goibhniu>
can you just run passwd from the store?
<goibhniu>
`echo /nix/store/*-passwd` should get you the path
<jophish>
hHi all
<MoreTea>
hi jophish
<jophish>
I have a package and I'd like to get a list of all the dependencies of this package, as well as their licensing information
<jophish>
getting the list of dependencies isn't a problem, but I'm not sure how to go from a store path to a license. As far as I can tell this isn't possible
<MoreTea>
You probably can write a nix expression to do that
<LnL>
for runtime dependencies you can use nix-store -qR $(nix-build -A hello)
<jophish>
LnL: yeah, that's what I've got so far :)
<LnL>
you might be able to get the licenses from the drv files
<cutuchiqueno>
goibhniu: since I have no globbing available that didn't work. However, I got it with `ls` and `grep`. Though there are more than one entries 2 with `.pam` and 2 with `*.pam.drv`.
ilja_kuklic has joined #nixos
<jophish>
LnL: I don't think it's in there, at least not in the couple I checked
<jophish>
MoreTea: changing mkDerivation or something so that the license ends up in the derivation?
<LnL>
nope doesn't look like it
<goibhniu>
cutuchiqueno: ah cool ... if you figure it out ... you'd be a hero if you add it to the manual :D
<MoreTea>
jophish, nope, you can access e.g. `pkgs.vim.meta.license`
<MoreTea>
I'm working on a nix expression to compute this information for you right now.
<jophish>
MoreTea: Then the problem becomes finding an attribute path for every dependency!
<jophish>
some might not even have one!
<MoreTea>
pkgs.vim.nativeBuildInputs seems to be about right
<MoreTea>
I'm not 100% sure, maybe someone else can comment on that?
<jophish>
That might not be the runtime dependencies, but that sounds like it might work
<jophish>
get that list, and then filter by the runtime dependencies
<MoreTea>
Hm. I'm actually not sure where Nix would store the runtime dependencies.
<MoreTea>
I know it does conservative scanning of the output files for /nix/store paths/hashes
<jophish>
yeah, buildInputs is not necessarily going to be a superset of the runtime dependencies
<LnL>
it's stored in the db
<MoreTea>
Yeah, but not in .drv files, right?
<MoreTea>
because those just contain the build instructions.
jensens has joined #nixos
<MoreTea>
In that sense, not shipping build dependencies is just an optimization
<LnL>
you can get the build inputs for a drv file, just like any other store path
<jophish>
perhaps modifying mkDerivation to put the json string of the meta information in the derivation would work then
<MoreTea>
@jophish, you can already access that through nix itself
<LnL>
but you'd have to create an expression that maps drvs to a license for each attribute in nixpkgs
<MoreTea>
I guess you need to script this in any case; there is no one command that will give you this answer, right?
<jophish>
that's for sure
<cutuchiqueno>
goibhniu: well, just using the passwd after doing the `export` command in the wiki did the trick. I am not sure what there is to add really?
<NixOS_GitHub>
[nixpkgs] fpletz pushed 1 new commit to master: https://git.io/vM5kF
<NixOS_GitHub>
nixpkgs/master e10cd27 Franz Pletz: Merge branch 'staging'
<goibhniu>
cutuchiqueno: great that you got it! .... the thing is, the wiki is frozen, so we can't edit it there ... and we don't have this in the manual yet (AFAICS) ... so explaining how to find the passwd command path etc. in the manual would be awesome
<cutuchiqueno>
goibhniu: Ok, I will look into it later on. Looking really forward to get to learn NixOS. Thanks for the help.
<goibhniu>
fantastic, cheers!
<NixOS_GitHub>
[nixpkgs] lsix pushed 3 new commits to master: https://git.io/vM5I2
<NixOS_GitHub>
[nixpkgs] mdaiter opened pull request #22049: scotch: specify flex version (master...scotchWork) https://git.io/vM5Lj
msd has joined #nixos
<msd>
Hey! Can anyone please check this? https://github.com/NixOS/nixpkgs/pull/22049 <- very simple push, just to specify the version of flex when compiling a specific package
<MoreTea>
jophish, interesting puzzle. I've managed to grab some information, but it's missing out on what apperas to be some stdenv stuff
<jophish>
MoreTea: something to start with, thanks a lot!
<MoreTea>
Not sure how gawk and glibc ends up in there.
<MoreTea>
Well, obviously you'd need some glibc, but I meant to say that I'm not sure where it originates from.
<MoreTea>
The mapping of `derivation` in Nix to a .drv is injective, and I don't think that license information is stored in .drv files.
<MoreTea>
So the only way to access that information is to either add the license info to a .drv file, or to find a way to traverse the runtime dependencies via nix.
<jophish>
yeah, it's not. It shouldn't be too hard to get it in there though with a different mkDerivation
<MoreTea>
The argument for not having licence information is probably storage bloat
<jophish>
It's probably not very often someone has to do this
<MoreTea>
I guess this could be solved by running some shell script in Nix.
<MoreTea>
I'll try that next.
<jophish>
well, can't one change mkDerivation purely in nix
<MoreTea>
hmm. no wait, I can't go back from from a .drv to a nixpkg path
<jophish>
make a new stdenv with a different mkDerivation which makes some dummy environment variable containing the json representation
<MoreTea>
That will force you to recompile everything
<jophish>
I don't think so, I don't need the derivations built
<jophish>
unless I have some IFD stuff going on
<jophish>
which I do :(
<MoreTea>
you do, otherwise the runtime dependency information cannot be found in the DB.
<jophish>
building everything again isn't the end of the world
<jophish>
oh yeah, you're right :)
<MoreTea>
Those dependencies are found by scanning the result of the build derivation.
<MoreTea>
*built derivation
<jophish>
I could get a superset of that info
pie_ has joined #nixos
<MoreTea>
That's what I tried. If you're OK with an apporoximation, you could probably just parse the names in the output of `nix-store -qR $pkgname` and try to find them in nixpkgs
proteusguy has quit [(Remote host closed the connection)]
<clever>
this inserts a bit of bash script that gets ran before the initrd tries to bring LVM online
<clever>
and because it references a part of iscsi, the iscsi package is included into the initrd automaticaly
<clever>
line 56 then connects to iscsi drives, allowing the existing stuff to find a rootfs over that
<clever>
lassulus: but it also depends on what kind of file you want to add, and why
<lassulus>
I wanted to add a keyfile to encrypt my luks encrypted device
lsix has joined #nixos
<clever>
ah
<lassulus>
So I don't need to input my password twice
hlavaty has joined #nixos
<clever>
but then anybody that can read /boot can get the keyfile and there goes the whole point of luks
<lassulus>
encrypt / decrypt
<clever>
enless you have /boot also encrypted?
<lassulus>
/boot is encrypted
<clever>
ah
<lassulus>
I'm using loader.grub.enableCryptodisk = true;
<clever>
if you use the keyfile path unquoted in configuration.nix, it will just happen automagicaly
<clever>
so boot.initrd.luks.device.foo.keyFile = ./keyfile.bin;
<lassulus>
but keyFile needs to be a string?
<clever>
any time you put a raw path in like that, nix will import the file to /nix/store, and convert the entry into its /nix/store path, as a string
<clever>
and the file becomes world-readable
<clever>
and now the keyfile is a dependency of your luks config, and just gets added to the initrd automaticaly
<hlavaty>
emacs is broken on unstable: /nix/store/5ihhl8bsccambkc6phhsd4vmi3gka00j-stdenv/setup: line 886: /nix/store/89cab82wdiyii22cxmf50f88vc1wj3y0-emacs-25.1/bin/emacs: cannot execute binary file: Exec format error
<MoreTea>
this might not work if you have older generations installed
<clever>
lassulus: "simplest" safe option i can think of on the spot, is to put the rootfs keyfile on /boot/, and then using something similar to my iscsi stuff, mount /boot before /, read the keyfile out, then umount /boot
<clever>
lassulus: this keeps the keyfile directly on /boot and unmanaged by nix
<clever>
so you can make it root-only
<lassulus>
well currently I don't have a seperate /boot partition
<clever>
ahh
<dweller>
or just copy it using initrd.extraUtilsCommands and add keyfile to device via initrd.luks.devices
<clever>
so you have a single / partition, and the grub decrypts it once via passphrase, then linux needs it again
<lassulus>
exactly
<clever>
dweller: thats similar to my original answer, and now the key is in /nix/store for all to see
<MoreTea>
It basically extracts the name from the /nix/store/abcdef-$NAME-$VERSION, and then fetches the license info from the nix-side. If the output path of the found package matches the original one, from which we guessed the package name, it's printed out. If it did not find the correct storage path, it aborts.
<jophish>
thanks MoreTea!
<lassulus>
my / is also in a lvm, but thats not important
<MoreTea>
It might help if you run this in a fresh VM, but will fail in any case if your package uses some non-standard dependency
<clever>
lassulus: similar to my laptop, on the top level i have a plaintext /boot, and a luks partition
<clever>
lassulus: luks then contains lvm, and one of the LV's contains ZFS
matthewbauer has quit [(Ping timeout: 240 seconds)]
<lassulus>
yeah, I just wanted to get rid of the unencrypted /boot
<MoreTea>
like stdenv.mkDerivation { buildInputs = [ (callPackage ../ffmpeg { enableSomeNonStandardFlag = true; }) ]}, since those are quite hard to discover
<jophish>
MoreTea: I might have to, I have so many old generations installed :)
<clever>
lassulus: related to my 2nd answer above, make 2 partitions, an encrypted /boot, and an encrypted /
<clever>
but that requires a reinstall
<lassulus>
np
arianvp2 has quit [(Ping timeout: 260 seconds)]
<lassulus>
I'm reinstalling like 3 times a day
<MoreTea>
jophish, yeah, it's an interesting question w.r.t. license management
<clever>
heh
<MoreTea>
It reminds me a bit of the work that the flying circus has done with Vulnix
<lassulus>
but I'm not sure why linux would be able to read /boot when its not able to read /
deepfire has quit [(Ping timeout: 255 seconds)]
<jophish>
MoreTea: I'll let you know when I get the perfect solution :)
<clever>
lassulus: oh right
Lowl3v3l has joined #nixos
<MoreTea>
you've got my email address if i'm not online here ;)
<lassulus>
I will try something with initrd.extraUtilsCommands
<jophish>
on another topic: is anything terrible going to happen if root uses NIX_REMOTE=daemon?
<clever>
lassulus: there are some more advanced things that might apply
<jophish>
MoreTea: I do?
<MoreTea>
it's in the latest gist :P
<clever>
lassulus: some bootloaders (possibly including grub) can append a 2nd CPIO archive to the initrd
<MoreTea>
(C) header
<jophish>
oh yes, thanks
<clever>
lassulus: so you could have a 2nd initrd file, root-only, with just the keyfile, that grub will read from /boot and append to the initrd
<clever>
then the key is in the initrd only at runtime
<clever>
its usualy triggered by just setting the initrd path twice
<clever>
but i havent confirmed this feature in grub
<MoreTea>
(by default, I like to release stuff under GPL-3, if that's a problem, I'll re-license)
<clever>
lassulus: and its 7am, i should get to bed, good luck
<lassulus>
clever: thank you, and sleep well
<jophish>
MoreTea: not at all, it's no problem
MP2E has quit [(Quit: sleep)]
deepfire has joined #nixos
Wizek has quit [(Ping timeout: 258 seconds)]
msd has quit [(Remote host closed the connection)]
msd has joined #nixos
deepfire has quit [(Ping timeout: 264 seconds)]
hlavaty has quit [(Remote host closed the connection)]
joelmo has joined #nixos
Wizek has joined #nixos
msd has quit [(Remote host closed the connection)]
<NixOS_GitHub>
[nixpkgs] canndrew opened pull request #22054: trezord: init at 1.2.0 (master...trezord) https://git.io/vM5uL
<garbas>
gchristensen: yes, but tomorrow. i want 1days twitter-buffer-zone to catch all the typos :)
<garbas>
1day
<gchristensen>
garbas: :) how about a label on github for interesting issues / PRs called "weekly"?
<binarin>
Hi. Is there a way to preserve a closure of everything that was used for `nixos-rebuild`? Otherwise after GC it starts re-downloading sources of customized packages, and requires manually re-adding things like jdk archive.
<ikwildrpepper>
garbas: can't find anything on ifttt.com about rss
<ikwildrpepper>
ah feed, damn
<ikwildrpepper>
garbas: what feed should I use?
<garbas>
gchristensen: +1 that would make the job easier
<MoreTea>
Now that @niksnut @ikwildrpepper and @garbas are all here, can I ask if there is interest in having "tutorials" or "guides" on nixos.org, that cut through all the abstractions of nix{,pkgs,os,ops}? It should heavily link back into the (reference) manuals.
indi_ has joined #nixos
<MoreTea>
These guides should be task based, showing people how to package their application, deploy it as a service on their machines, and then continue with how to deploy that service on a cluster of machines with nixops.
<garbas>
MoreTea: i would say build it as a separate site and once it is mature/ready we can include it into nixos.org
<gchristensen>
MoreTea: joepie91 might be interested in helping too
<NixOS_GitHub>
[nixpkgs] fpletz pushed 1 new commit to master: https://git.io/vM5w6
<NixOS_GitHub>
nixpkgs/master 33f4728 Franz Pletz: youtubeDL: 2017.01.18 -> 2017.01.22
* joepie91
appears
<MoreTea>
The point is, I'm only willing to invest time in this, if this is eventually going to be merged in nixos.org.
msd has quit [(Ping timeout: 255 seconds)]
<MoreTea>
So hence me asking if that is a possibility :)
<joepie91>
so I haven't finished my post about documentation yet, but the tl;dr conclusion/proposal of it is "do away with the reference guide, write more about things on a *conceptual* level (rather than just instructions), and make things more consistently documented"
<joepie91>
("do away with" in a structural sense; the single-page wall-of-text format doesn't work)
<NixOS_GitHub>
[nixpkgs] shlevy opened pull request #22057: (haskellPackages.callPackage foo).env: Set the right env vars when cross-compiling (master...haskell-modules-cross-env) https://git.io/vM5wb
<joepie91>
I just still need to write the explanations and examples that lead up to those points :P
<joepie91>
or part of them anyway
sphalerite has quit [(Ping timeout: 240 seconds)]
<Yaniel>
could someone give me a hint on packaging vulkan layers?
<joepie91>
also, gchristensen, re: the Tahoe guide... I document random stuff with some regularity :P
<NixOS_GitHub>
[nixpkgs] shlevy pushed 1 new commit to master: https://git.io/vM5rY
<NixOS_GitHub>
nixpkgs/master b82d6b3 Shea Levy: ios-cross: Expose the SDK directory via ccCross
msd_ has quit [(Remote host closed the connection)]
<Yaniel>
I'm trying to write an expression for renderdoc and it wants to put some json files into /etc/vulkan
<Yaniel>
and I have no idea where I should tell it to put them on nixos
indi_ has joined #nixos
eacameron has joined #nixos
erictapen has joined #nixos
sphalerite has joined #nixos
<joepie91>
MoreTea: anyhow, if you're going to be working on such guides, my most critical recommendation won't be "don't just tell people how to do it, explain to them *why* that is how to do it"
<joepie91>
put it into context, make the individual steps reusable bits of knowledge/understanding
indi_ has quit [(Remote host closed the connection)]
indi_ has joined #nixos
<MoreTea>
yes, thanks.
<Yaniel>
joepie91: often people want to "just get it done" though
<joepie91>
Yaniel: that's not a reason not to include such information :)
<MoreTea>
@Yaniel, it should be a balance between explaining concepts and "shut up, and just tell me what to copy&paste"
<joepie91>
those people can ignore parts of the article as necessary to meet their goals
<Yaniel>
ideally there'd be side-by-side a TL;DR and a longer "why"
<joepie91>
as long as the conceptual explanations are clearly distinguishable from the "do this and that" parts, there shouldn't be an issue
<MoreTea>
yep, but that requires some discipline to write paragraphs of in-depth stuff vs the bare minimals
<Yaniel>
(for example the ToS of gog.com is done this way)
aj__ has quit [(Ping timeout: 258 seconds)]
<joepie91>
Yaniel: you can read that article as either 1) a step-by-step guide, 2) a conceptual explanation with examples, or 3) just infer the explanation from the examples without reading the text around it
<joepie91>
depending on how much background knowledge you already have and what format is easiest to read for you
acowley_away has quit [(Ping timeout: 240 seconds)]
<erictapen>
I recognized, that pkgs.gnome3.gnome-terminal settings can't be changed. I'm quite sure that the reason for this is the non-writable config file in the nixstore. Before sending a PR, I'd like to know which solution would be more likely to be accepted: Putting the conf file into a stateDir or making options for all the stuff defined in the configfile?
<gchristensen>
erictapen: it seems gnome-terminal should know how to read / write configs in ~/.config, no?
<MoreTea>
joepie91, thanks
<joepie91>
(as an aside, so far I'm quite happy about the responses in here to the documentation issue... much better than in Python-land, where the response could be more or less summarized as "go away and stop complaining")
acowley_away has joined #nixos
<gchristensen>
joepie91: python already took over the world, we're still trying :)
msd has quit [(Ping timeout: 258 seconds)]
derjohn_mob has joined #nixos
sphalerite has quit [(Ping timeout: 240 seconds)]
<joepie91>
heh
<joepie91>
fair point
<Profpatsch>
:exf Maybe a -> MaybeT IO a
<erictapen>
gchristensen: ok, maybe I simply used it the wrong way. Atm I don't have access to a nixos machine but when I installed it, it didn't work by default
<Profpatsch>
Wrong channel again …
sphalerite has joined #nixos
<erictapen>
gchristensen: when i did Rightklick -> Edit Profile, the config window didn't allow me to change any value
<gchristensen>
yikes
<erictapen>
but good to know, that it is supposed to be in ~/.config
<gchristensen>
well, I think
xadi has quit [(Quit: Leaving.)]
msd_ has quit [(Remote host closed the connection)]
matthewbauer has joined #nixos
mkoenig has joined #nixos
dramforever has joined #nixos
<mbrgm>
when I'm wrapping a binary with `makeWrapper`, the following strip command does not recognize the fileformat... what am I doing wrong?
<gchristensen>
nothing, the strip command doesn't need to recognize the file format
<gchristensen>
it is just a notice
lukego has joined #nixos
matthewbauer has quit [(Ping timeout: 276 seconds)]
<Profpatsch>
Hm, it’s probably a bad idea, since if someone were to push a package like that to hackage, it would silently be overwritten.
rly has quit [(Ping timeout: 264 seconds)]
<lukego_>
Is it reasonable to run lots of nix-build commands locally on my laptop, for interactive hacking, and use "distributed builds" to have _everything_ actually executed on another host? Or will this be painfully slow compared with ssh'ing to the other host and running nix-build "locally" there?
eacameron has quit [(Remote host closed the connection)]
lukego has quit [(Ping timeout: 260 seconds)]
eacameron has joined #nixos
<gchristensen>
lukego_: I'd say try it and see :) and then come back and tell us
<gchristensen>
oh :(
<mbrgm>
gchristensen: ok, the error actually is 'builder for ‘/nix/store/c3qbqr521j4cy9irfngw98ch8qhk7rjp-journalbeat-5.1.2.drv’ failed to produce output path ‘/nix/store/wv4in0jqck4m0p1nil39f3mmx6k774l8-journalbeat-5.1.2-bin’'
<lukego_>
gchristensen: that's what I am doing, but if somebody knows that this path ends in frustration and wasted time then it would be nice to get a heads-up :)
acowley_away has quit [(Ping timeout: 258 seconds)]
<Profpatsch>
jophish: Well, I’d like to respect the choice of the author.
<Profpatsch>
If the person doesn’t submit it to hackage, it probably has a reason.
<jophish>
lukego_: It depends on what the ratio between build time and transmission time is
<jophish>
Profpatsch: exactly :|
<Profpatsch>
And people will go to him if something doesn’t work.
<gchristensen>
lukego_: I think it depends if your remote is a fast builder / if you're transfering big amounts of data to it in order to build each time
<gchristensen>
lukego_: and if xfer time + build time on teh remote is faster than just build time locally
acowley_away has joined #nixos
ryantrinkle has quit [(Ping timeout: 245 seconds)]
<lukego_>
Sounds promising for my case.
<LnL>
distributed builds will also copy the paths to the host you are building from once it's finised
<lukego_>
LAN with wimpy laptop remoting to beefy server
<lukego_>
Or generally Mac machine remoting to Linux server
<joepie91>
is there a documented packaging process for Unity games?
<joepie91>
seems to use Mono
<gchristensen>
do we have any unity games?
<joepie91>
alternatively, an existing expression
<joepie91>
I have no clue :P
<joepie91>
quick source search didn't turn up any hits for "unity"
<sphalerite>
not sure about any games written with unity
<joepie91>
right, that's just the editor
<gchristensen>
lukego_: yeah, try it :)
<joepie91>
(nodejs? wut?)
* joepie91
suspects that this game will be a minor nightmare to package
acowley_away has quit [(Ping timeout: 264 seconds)]
<sphalerite>
Which game?
<joepie91>
RimWorld
rly has joined #nixos
iyzsong has quit [(Quit: bye.)]
<joepie91>
I think I'll defer packaging this for a while :)
mudri has joined #nixos
<sphalerite>
looks like a neat game
acowley_away has joined #nixos
<lukego_>
I'm getting "warning: the group ‘nixbld’ specified in ‘build-users-group’ does not exist" and then later: Died at /nix/store/hca2hqcvwncf23hiqyqgwbsdy8vvl9xv-nix-1.11.6/libexec/nix/build-remote.pl line 115, <STDIN> line 1. killing process 13295: Operation not permitted
<gchristensen>
Friday I was shocked to see how stinking well steam worked
mog has left #nixos ["Leaving"]
<rly>
Is there any job execution system which is open-source and allows for job state to be persisted across multiple nodes? I.e., is fault tolerant?
<lukego_>
Do remote builds "just work" when running on OSX? I installed nix with the curl command and created /etc/nix/remote-systems.conf
<gchristensen>
rly: chronos?
<joepie91>
rly: warning ahead of time; you will almost certainly have to settle for the possibility of jobs either occasionally getting lost or duplicated
<gchristensen>
lukego_: you should ping LnL :)
<joepie91>
(no matter what you use)
<LnL>
lukego_: pong
<mbrgm>
can someone tell me why this (http://nixpaste.lbr.uno/M4gXIaL3?nix) is giving me this error? builder for ‘/nix/store/ik1c6mc3k8l39rfpf40r34nsqbrv5zwr-journalbeat-5.1.2.drv’ failed to produce output path ‘/nix/store/gfx8ms7wxwbhh9afh4sp04wfdhbb96px-journalbeat-5.1.2-bin’
<lukego_>
ping pong?
<gchristensen>
joepie91: almost certainly? mathematically certain! :)
<mbrgm>
what needs to be done to produce the '-bin' path? when I drop the installPhase from the derivation, everything works as expected
ixxie has joined #nixos
<LnL>
lukego_: I wrote some instructions on how to setup distributed builds with a docker container
<lukego_>
ah :-)
<lukego_>
Just now I am keen to avoid docker but could be Plan B or C :)
<joepie91>
gchristensen: theoretically, I think it would be possible to never have any duplicated or lost jobs, so long as you're willing to trade in speed for that and allow a job to remain in "pending" indefinitely ;)
<joepie91>
but that's a completely unuseful tradeoff, heh
<lukego_>
LnL: Thanks for the tip. I only have ~ 30 minutes to spend on this atm so it seems clear I should find another solution for now.
<LnL>
lukego_: it's similar with a real machine, you don't have to use that container
<gchristensen>
joepie91: no, exactly once is impossible:http://bravenewgeek.com/you-cannot-have-exactly-once-delivery/
proteusguy has joined #nixos
<LnL>
lukego_: don't remember exactly in what cases you need to setup the signing key, might not be neccecary
<gchristensen>
sorry mbrgm
<gchristensen>
mbrgm: you're probably writing stuff to the directory represented by the variable $out, but you also need to write to the directory represented by the variable $bin, which will usually be $bin/bin/<yourbin>
cfricke has joined #nixos
<joepie91>
gchristensen: will have to read that when I'm fully awake
<joepie91>
:P
<gchristensen>
joepie91: the gist of it is how do you know it wasn't delivered but the ack of delivery was missed?
<lukego_>
LnL: ... though if this would somehow work in single-user mode that would be awesome for me.
<rly>
joepie91: I basically need something which has an uptime of 99.99% and doesn't run things twice, etc.
<mbrgm>
gchristensen: ahh, thank you. when I write to the $bin path in addition, when does nix use the contents of $out?
<joepie91>
rly: see gchristensen's link :P
<LnL>
lukego_: it should both work work for both a single user install and with the nix-daemon
<rly>
joepie91: I.e., something which actually works, not just something someone made at some point.
<LnL>
lukego_: might be that you need to setup the signing key with a single user install
<joepie91>
gchristensen: task runner can acknowledge completion, task router can acknowledge acknowledgment, task runner can periodically re-attempt notification of completion until it has received ack-ack from task router, in which case your task is always in either a "confirmed and done" or "not confirmed... YET" state
<gchristensen>
rly: dkron and chronos are legit
<joepie91>
gchristensen: but like I said, this is not a useful tradeoff or implementation :P
<joepie91>
it's very much a "technically..." argument, heh
<gchristensen>
joepie91: ok so the task runner gets the job says ok I got the job then does the job then goes away forever. the job is now done. does your task runner now assume it wasn't done and retries (done again) or assumes it was done and doesn't retry?
<rly>
Exactly once is possible in our universe.
<gchristensen>
joepie91: it never says "ok I did the job"
<rly>
Exactly once in some theoretical universe with certain assumptions is not.
<joepie91>
gchristensen: task router does not attempt to send the job to another worker, but rather indefinitely waits for the worker to return
<joepie91>
to re-send the job, which the worker can locally recognize as "I already have this"
<joepie91>
ie. a job is tied to a specific worker
<gchristensen>
joepie91: ok so same situation except the task runner ceases to exist before completing the job, this means it never retries thusly the task is nevre done
<joepie91>
(without that constraint it can indeed not work)
<joepie91>
gchristensen: my argument is based on the assumption that a task runner is eventually fixed and returned
<joepie91>
ie. there is no point at which a task runner "no longer exists", it is just temporarily out of commission, for potentially very long values of "temporarily"
<joepie91>
hence why there's no delivery time guarantees
msd has quit [(Read error: Connection reset by peer)]
<gchristensen>
joepie91: we could talk about this a long time, but it is off topic for #nixos... but I promise exactly once delivery is impossible
msd has joined #nixos
<joepie91>
hehe
<joepie91>
I should probably get to work anyway
<joepie91>
:p
<gchristensen>
we could go to PM if you'd like me to explain further
<joepie91>
gchristensen: probably better for a later time when I have less on my plate :D
<gchristensen>
ok, in the mean time, assume it is impossible
<joepie91>
gchristensen: I agree that for all practical purposes it's impossible, this was just a theoretical argument :P
<gchristensen>
but that is wrog
dramforever has quit [(Ping timeout: 240 seconds)]
<erictapen>
ok, I had another look into the issue of unchangable config in gnome-terminal. It looks like, that gnome-terminal isn't using a conf file anymore, but it uses dconf. Does anyone know of a good method for handling dconf config?
msd_ has quit [(Client Quit)]
msd has quit [(Ping timeout: 256 seconds)]
<lukego_>
How do dependencies work with remote builds? If I tell nix to run builds on a remote machine then will that machine download the dependencies directly, or will my local machine download all of the sources and copy them to the remote machine for building one at a time?
sphalerite has quit [(Remote host closed the connection)]
<gchristensen>
lukego_: I believe your machine will download, then xfer to the remote
<lukego_>
OK. So could be that I would be better off shipping the nix expression to the remote host and copying the closure of the result instead.
<gchristensen>
well
<gchristensen>
maybe
<gchristensen>
it'll only transfer dependencies if they're not already t here
<lukego_>
ok. to be continued... :)
<lukego_>
Thanks for the tips all!
<gchristensen>
good luck! you're welcome!
erictapen has quit [(Ping timeout: 256 seconds)]
ixxie has quit [(Remote host closed the connection)]
sitedyno has quit [(Ping timeout: 240 seconds)]
SovereignBleak has joined #nixos
<NixOS_GitHub>
[nixpkgs] jgeerds pushed 2 new commits to master: https://git.io/vM5yi
<NixOS_GitHub>
nixpkgs/master 17af9e9 Peter Hoeg: terragrunt: 0.9.1 -> 0.9.3
<jkemet>
Hi! Could any of you provide me with some insights on using `nix-shell` for developing cmake based C++ projects?
acowley_away has joined #nixos
estewei has joined #nixos
estewei has quit [(Client Quit)]
<domenkozar>
jophish: best to ash shlevy
<Yaniel>
Sure, any specific question?
<domenkozar>
ask*
<domenkozar>
lol
estewei has joined #nixos
<gchristensen>
man I don't think that will work, and that is very silly
<jkemet>
I have been using `nix-shell` with a shell.nix file for quite some time now, but I just realized that my approach seems flawed. I have had good success with defining the dependencies of my project using `buildInputs`. Now I tried to use the same shell.nix file on a different machine, but I kept having problems with "error while loading shared libraries:".
<lassulus>
have you tried it with nix-shell --pure?
<lassulus>
on the machine where its working
<jkemet>
So now I realized that this was only working, because I have gcc-wrapper installed.
<jkemet>
And cmake is using libstdc++ from the `nix-env -iA gcc`, which I have also in my LD_RUN_PATH.
matthewbauer has quit [(Ping timeout: 255 seconds)]
<jkemet>
So I can get everything to compile, but the output is missing a libstdc++ runpath baked into it.
<jkemet>
It all ends up fine if I use `nix-build`, but inside the `nix-shell` the linking of libstdc++ just doesn't work correctly.
<Mic92>
gchristensen: I am jealous of your new shiny private hydra. Do you have some automation to schedule it for pull requests?
systemfault has joined #nixos
<gchristensen>
Mic92: PRs are automatically added to the hydra the moment they're created via a web-hook, and then an evaluation is triggered every time it is updated :)
<gchristensen>
fpletz: I thought flexget didn't exist on stable?
Lowl3v3l has quit [(Ping timeout: 240 seconds)]
erictapen has joined #nixos
Lowl3v3l has joined #nixos
asseadinho has joined #nixos
<gchristensen>
garbas: Lobeste.rs says we should call it the NixOS Fortnightly >.>
pi3r has quit [(Ping timeout: 252 seconds)]
seppellll has joined #nixos
erictapen has quit [(Ping timeout: 240 seconds)]
tilpner has joined #nixos
<seppellll>
hi, does anybody have a pointer to a good read on "system testing" with nixos? maybe i am just to stupid to search but i cannot find anything on google.
bfrog_ has joined #nixos
<LnL>
what do you mean by that?
xadi has joined #nixos
osa1 has joined #nixos
flvzsch50 has joined #nixos
<osa1>
hi all. can I run deploy a nixops spec on a non-nixos Linux machine?
<seppellll>
people keep talking about how you can test your applications with nix in a distributed/network scenario. specifying your machines nixos style, defining test scripts for each machine and then spin up some VM to test your application. unfortunately the nix manual does not mention this in any way
JagaJaga has joined #nixos
<gchristensen>
oh
<gchristensen>
seppellll: check the nixos manual and the nixos/tests directory
<seppellll>
i was such a derp not to check the nixos manual. thx gchristensen
<gchristensen>
:)
simendsjo has joined #nixos
<jkemet>
@lassulus: I did now try to create a minimal working example with `nix-shell --pure` and it seems to work that way. That's good. I thought this workflow might not be supported after all, but it's simply that my environment variables seem to mess up the build without --pure.
<jophish>
Is this something which can be bumped to the failing package itself, why didn't the tests break?
athan has joined #nixos
<gchristensen>
sorry?
<jophish>
ah, no tests
<gchristensen>
ah
<jophish>
or better yet, the failing package wouldn't even type check!
modafinil has joined #nixos
georges-duperon has joined #nixos
teknico has left #nixos []
rly has quit [(Ping timeout: 248 seconds)]
asymmetric has joined #nixos
goibhniu has quit [(Ping timeout: 256 seconds)]
<LnL>
is it evaluated by hydra?
<mbrgm>
when using 'nox-review ...', is it mandatory the nixpkgs repo lives in ~/.nox/... ? can I use my regular working copy, which lives at a different path?
derjohn_mob has quit [(Ping timeout: 264 seconds)]
<shanemikel>
hey LnL , for that script, is there any reason to trap all those signals? I thought EXIT takes care of everything
<LnL>
not sure, could be
AllanEspinosa has joined #nixos
mudri has quit [(Ping timeout: 256 seconds)]
<NixOS_GitHub>
[nixpkgs] rycee pushed 3 new commits to master: https://git.io/vMdSR
<NixOS_GitHub>
nixpkgs/master 3c521ed Robert Helgesson: svtplay-dl: 1.8 -> 1.9
<NixOS_GitHub>
nixpkgs/master 653a7d9 Robert Helgesson: perl-MooseX-Types-Common: 0.001013 -> 0.001014...
<NixOS_GitHub>
nixpkgs/master df89b9b Robert Helgesson: perl-MooX-Types-MooseLike-Numeric: 1.02 -> 1.03...
athan has quit [(Read error: Connection reset by peer)]
sitedyno has quit [(Ping timeout: 240 seconds)]
tg has quit [(Excess Flood)]
athan has joined #nixos
JagaJaga has joined #nixos
AllanEspinosa has quit [(Ping timeout: 240 seconds)]
osa1 has joined #nixos
tg has joined #nixos
athan has quit [(Ping timeout: 240 seconds)]
athan has joined #nixos
athan has quit [(Client Quit)]
eacamero_ has quit [(Read error: Connection reset by peer)]
eacameron has joined #nixos
athan has joined #nixos
athan has quit [(Client Quit)]
mexisme2 has joined #nixos
<NixOS_GitHub>
[nixpkgs] LnL7 opened pull request #22072: kwm: init at 4.0.4 (master...kwm) https://git.io/vMd58
sitedyno has joined #nixos
morsicus1 has joined #nixos
_rht has joined #nixos
osa1 has quit [(Ping timeout: 252 seconds)]
mexisme has quit [(Ping timeout: 248 seconds)]
p0le has quit [(Quit: WeeChat 1.6)]
athan has joined #nixos
gciolli has quit [(Remote host closed the connection)]
xadi has joined #nixos
morsicus1 has quit [(Quit: Leaving.)]
morsicus1 has joined #nixos
simendsjo has quit [(Quit: ERC (IRC client for Emacs 24.5.2))]
AllanEspinosa has joined #nixos
seanz has quit [(Quit: Leaving.)]
<NixOS_GitHub>
[nixpkgs] romildo opened pull request #22073: Upd.pcmanfm (master...upd.pcmanfm) https://git.io/vMdbd
morsicus1 has quit [(Quit: Leaving.)]
Itkovian has joined #nixos
jmiven has quit [(Quit: co'o)]
Itkovian has quit [(Quit: My MacBook has gone to sleep. ZZZzzz…)]
jmiven has joined #nixos
Itkovian has joined #nixos
xadi has quit [(Quit: Leaving.)]
xadi has joined #nixos
justbeingglad has joined #nixos
justbeingglad has left #nixos []
morsicus1 has joined #nixos
zahary has joined #nixos
Guest74060 has joined #nixos
<NixOS_GitHub>
[nixpkgs] basvandijk opened pull request #22074: opencv-3.1: get ippicv from the nix store (master...opencv-ippicv-no-download-fix) https://git.io/vMdxW
Itkovian has quit [(Quit: My MacBook has gone to sleep. ZZZzzz…)]
xadi has quit [(Quit: Leaving.)]
Guest74060 has left #nixos ["Once you know what it is you want to be true, instinct is a very useful device for enabling you to know that it is"]
<zahary>
I'd like to have a Nix shell where a local copy of boost is built using an older version of clang. can you give me some pointers on how to approach this? currently, I set my buildInputs to [ llvmPackages_37.clang boost160 ]; but this will install a binary distribution of boost created using gcc. later, this leads to a linker errors, because the gcc build cannot be used with clang
<zahary>
what do I need to override/replace in order to have the boost expression produce a local build using clang?
<NixOS_GitHub>
[nixpkgs] LnL7 pushed 2 new commits to staging: https://git.io/vMdpR
<NixOS_GitHub>
nixpkgs/staging 02341c2 Daiderd Jordan: db: disable dbm support by default
<NixOS_GitHub>
[nixpkgs] bjornfor pushed 1 new commit to master: https://git.io/vMdht
<NixOS_GitHub>
nixpkgs/master 1853f17 Bas van Dijk: opencv-3.1: get ippicv from the nix store...
<NixOS_GitHub>
[nixpkgs] bjornfor closed pull request #22074: opencv-3.1: get ippicv from the nix store (master...opencv-ippicv-no-download-fix) https://git.io/vMdxW
seanz has joined #nixos
athan has quit [(Ping timeout: 260 seconds)]
<slyfox>
zahary: i think you need something similar to 'accelio = callPackage ../development/libraries/accelio { stdenv = overrideCC stdenv gcc5; };'
<nekroze>
Is there anything for deploying nix closures/builds with ansible?
cfoch has quit [(Remote host closed the connection)]
<cransom>
nothing that i'm familiar with. though, rolling your own module would be just 2 plays, no? send configuration.nix, run nixos-rebuild switch.
johnsonav has quit [(Ping timeout: 255 seconds)]
<nekroze>
this is for non nixos hosts. Just want to package software with nix as a package manager. aparently using ssh to send the closure in a script is not portable enough and there are no modules out there for ansible to handle it
shanemikel has quit [(Quit: My MacBook Air has gone to sleep. ZZZzzz…)]
hiratara has joined #nixos
xadi has joined #nixos
Itkovian has quit [(Ping timeout: 240 seconds)]
mexisme2 has quit [(Ping timeout: 255 seconds)]
Itkovian has joined #nixos
<LnL>
nekroze: you can build a tarball like this nix-store -qR <path> | tar -czf build.tar.gz -T -
phreedom has quit [(Remote host closed the connection)]
<makefu>
hi, i wanted to use pypi2nix for packaging grr-incident-client but it somehow fails with ImportError: No module named six.moves ... anyone has seen something like this before? Output is https://paste.krebsco.de/CzTrzd9P
Itkovian has quit [(Quit: My MacBook has gone to sleep. ZZZzzz…)]
<makefu>
i also tried to add the missing dep with -E python27Packages.six but no change